LANDSCAPE DYNAMICS ANALYSIS OF GUIDE WETLANDS
IN YELLOW RIVER WATERSHED BY LANDSAT SERIES DATA
Xuying Shi; Fanghua Hao*, Cheng Zhao, Dan Wang, Wei Ouyang
School of Environment, State Key Laboratory of Water Environment Simulation, Beijing Normal University, Beijing,
100875, China
WG II/2 - Spatial Reasoning, Analysis, and Data Mining
KEY WORDS: Image interpretation; Landuse; Terrestrial photogrammetry; Spatial-temporal analysis; Dynamic Change; Hydrology;
Wetlands; Yellow River
ABSTRACT：
Wetlands are resources of paramount importance with many values and functions. However, with the economy development and the
population growth, wetlands have undergone great changes. This study investigated the land-cover and landscape pattern dynamics
of Guide wetlands in the periods 1977-2006. Four Landsat images were used to locate and quantify the changes. The study revealed
the area of wetlands previewed a downward trend during the past 30 years, whereas the area of cultivated land increased all the time.
From 1997 to 2000, the wetland landscape pattern became more fragile and less connective. The landscape types converted to each
other dramatically. During 2000 to 2006, both dominant and contagion indices on the whole area were enhanced. Its driving forces
were analyzed according to socioeconomic development and climatic information. Remote sensing (RS) and Geographic information
system (GIS) technologies have proved to be useful tools for assisting decision-makers to locate and quantify changes in land
resources, and hence to identify appropriate solutions for sustainable management of wetlands.
1. INTRODUCTION
Wetlands are in the amphibious staggered zone of transition,
which comprise about three to six percent of the earth's land
surface, but they have various ecology function, economy value
and society value, including agricultural production, fisheries,
provision of wildlife habitat and so on (Acreman and Hollis,
1996 ； Sugumaran et al., 2004). However, with the
socioeconomic development and the human activities
reinforcement, wetland resources have been seriously disturbed,
especially the large numbers of reservoirs are constructed on the
Yellow River, which greatly impacts the wetland hydrological
process and characteristics. Therefore, the study of dynamics in
Guide wetlands in Yellow river watershed has an important
significance for the wetland protection and local environmental
management.
Baker et al., 1991). The objective of this study is to investigate
the dynamics of Guide wetlands through the analysis of changes
in land-cover and landscape pattern from 1977 to 2006, and the
impacts of human activity on wetlands by utilizing RS and GIS
techniques.
2. MATERIALS AND METHODS
2.1 Description of the study area
After the construction of reservoirs, the land-cover and
landscape pattern along the Yellow river changed obviously.
Many researchers focus on a lot of studies associated to
sediment, flood, and runoff of the Yellow River (Huang, 2002;
Xu, 2002; Jiang, 2008). Also many scholars study the wetland
ecosystem and landscape dynamics of the Yellow River Delta
(Li, 2007; Yoshiki Saitoa, 2000; Cai, 2006). However,
relatively less attention has been paid to studying the wetland
landscape and land-cover change along the Yellow River and
the factors affecting the change. Therefore, elucidation of the
dynamic change of Guide wetlands is essential for wetland
management.
With rapid changes in landscape occurring over large areas,
remote sensing has become an essential tool for monitoring
such changes, particularly as a means of complementing or
updating conventional data gathering techniques (Nellis, 1986;
* Corresponding author.
author.
Figure1. Location of the Guide wetlands
This is useful to know for communication with the appropriate person in cases with more than one
247
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B2. Beijing 2008
The study area is located in Guide country (N35°30'～36°30',
E101 ° 0' ～ 101 ° 50'), Qinghai province, with an area of
approximately 6028ha (Figure1).
The Guide wetlands are joined by a narrow band of land along
the Yellow River, which situated between the Longyangxia
reservoir and the Lijiaxia reservoir. This region belongs to the
transition zone between Qinghai-Tibetan Plateau and Loess
Plateau. The feature of climate is the typical alpine continental
climate with hyper-arid natural condition. Mean annual solar
radiation is 2928h and yearly mean air temperature is about
7.2℃. The coldest and warmest average annual temperature is
-23.8℃ in winter and 34℃ in summer. Rainfall is highly
variable between 251-599mm per year. The study area includes
78.8km long stretches of the Yellow River and has abundant
wetland types such as shrub wetland, reed marsh and meadows
wetland. With the population growth and economy development,
the human being and nature have put wetlands under the
enormous pressure. Vast areas of wetlands continue to be
converted to cultivation and grazing, both of the land-cover and
the landscape pattern have undergone great changes. This paper
therefore examines the dynamics of Guide wetlands and their
implications.
2.2 Data and land cover classification
The wetlands dynamic information was extracted from the
interpretation of RS data in 1977, 1994, 2000 and 2006. Table 1
presents the satellite imagery input data used for study. The
Landsat images were enhanced using the linear contrast
stretching and histogram equalization to a common ALBERS
coordinate system based on 1:50000 topographic maps of China.
With the aid of Global Positioning System (GPS) equipment,
maps and key informants, the various land-cover types were
located in the field and their positions recorded. The
unsupervised classification with the maximum likelihood
method algorithm was conducted to classify the image
Index
LSI
containing the study area. Combined with field survey, the
following major land-cover categories were identified:
cultivated land (CL), watershed (WA), tidal flat (TF), reed
marsh (RM), meadows wetland (MW) and shrub wetland (SW).
Image
Landsat MSS
Landsat TM
Landsat TM
Landsat ETM
Path/Row
142/35
132/35
132/35
132/35
Format
GeoTIFF
GeoTIFF
GeoTIFF
GeoTIFF
2.3 Methods
The magnitude and direction of changes in landscape are the
most important factors relating to landscape evolution (Antrop,
2000). With the wetlands information of land-cover types from
the four images, land-cover transition matrices were elaborated
for the three periods by using Overlay Tool in the GIS software.
Each matrix represents either the probability of persistence of
the period, or the probabilities of transition to another
land-cover category during the same period (Alejandro, 2007).
The Patch Analyst module was used to calculate the landscape
metrics, which is an extension to the ArcView GIS system that
facilitates the spatial analysis of landscape patches, and
modeling of attributes associated with patches (Rempel and
Carr, 2003). By FRAGSTATS, two levels of metrics were
computed. i.e., class level, which means each land use type in
the landscape mosaic, and landscape level, which means the
landscape mosaic as a whole (Lu et al., 2003). In order to detect
landscape change of the study area, seven indices were analyzed:
patch density (PD), mean patch area (AREA_MN), landscape
shape index (LSI), contagion index (CONTAG), dominance
index (D), Shannon's evenness index (SHEI) and Shannon's
Diversity Index (SHDI) (Table 2).
Mathematic model
Ecological significance
Landscape shape index indicates the
complexity of shape.
LSI = 0.25E / A
Shannon’s diversity index is a popular
measure of diversity in community ecology,
applied here to landscapes.
m
− ∑ ( Pi ln Pi )
SHDI =
Season
Wet
Wet
Wet
Wet
Table1 Landsat images applied in analysis
E is length of total patches borderlines; A is the total area.
SHDI
Year
1977-07-15
1994-08-16
2000-07-05
2006-08-05
i =1
Pi is proportion of the landscape occupied by patch type (class) i ;
m is number of patch types (classes) present in the landscape.
−
SHEI
SHEI
D
=
m
∑ (P
i
i =1
Shannon’s evenness index is expressed such
that an even distribution of area among patch
types results in maximum evenness.
ln Pi )
ln m
m
D = H max + ∑ (Pi ) × log 2 (Pi )
i =1
,
H
max
= log
2
(m )
Dominance index indicates one or fewer
dominant classes exist in the landscape
Table2 Models and significances of landscape indices
3. RESULTS AND DISCUSSION
3.1 The analysis of wetland land cover change
In general, the results show that only the CL area increased
noticeably during 30 years. The SW area had no obvious change,
while the areas of other landscape types all registered a
descending trend. The WA area which occupied 2479ha in 1977,
248
and decreased to 1417ha in 1994, was the most remarkable
change, and the reed marsh was the second (Figure2). Based on
the four-year land-cover information analysis, the land-cover
transition matrices for the periods of 1977-1994, 1994-2000 and
2000-2006 were obtained (Table3-5). During the period
1977-1994, both RW and WA areas dropped dramatically. The
annual average changing rates of RW and WA reached -4.22
and -2.53, respectively. The cause of decrease in RW and WA
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B2. Beijing 2008
area was changing into MW, TF and CL. It demonstrated that
the water source was scarce and human beings disturbed
wetlands frequently. From 1994 to 2000, except TF, the
changing rates of other types were all positive values. The area
of TF was mainly converted into RM (10.48%) and WA
(33.95%). The changing rate of CL was 0.35 compared with
8.83 for 1977-1994. It indicated that the water source of
wetlands was supplied in time, and the trend of changing
wetland to CL decreased. During 2000-2006, the CL area
increased remarkably from 30.76% to 40.36%. The expansion
of CL mainly resulted from shrinkage of MW (20.43%), WA
(3.78%) and TF (4.77%). It previewed that the human beings
reclaimed farmland from the wetlands dramatically.
3.2 The analysis of landscape pattern dynamics
All these land use changes affected landscape patterns. The
calculated results of landscape indices were listed in Table 6
and Table7. The results show that the values of SHEI and SHDI
had nearly no change during 30 years. Both of them only
dropped lightly after 2000. It demonstrated that the landscape
heterogeneity decreased, and the distribution of land-cover
became uneven after 2000. The dominance index had a certain
extent rise, which resulted from the area of CL increased
remarkably and became the dominant landscape type.
The PD of wetlands on the whole area increased from 1977 to
2000, while decreased from 2000 to 2006. The CONTAG had a
converse trend compared with the PD. It showed that the
wetlands became more fragmented during 1997-2000, whereas
trended to concentration after 2000. The PD of RM, WM and
CL had the same trend with the whole landscape. During
2000-2006, the PD of CL and WA decreased while the
AREA-MN of them was on an ascending trend. The
fragmentation decreased on the whole mainly because that CL
extended to connect each other.
3000
RM
WA
2500
SW
TF
MW
CL
Area (ha)
2000
1500
1000
Except TF, the LSI of other land covers all showed a rise trend
from 1977 to 1996, the LSI of WA which reached 7.0112 in
1996, was the most obvious change. It indicated that the shape
of patches inclined to complicated and irregular. The main
reason was that the landscape types converted to each other
dramatically. The LSI of CL was on a descending trend after
1996, which was mainly attributable to the area of CL increased
continually and the shape of CL patches inclined to regular and
simple.
500
0
1975
1980
1985
1990
1995
2000
Year (a)
2005
2010
Figure2. Total area of different land covers from 1977 to 2006
1994
1977
RM
SW
MW
WA
TF
CL
Toatl1994
PL (%)
AACR (%)
RM
17
0
21
117
8
11
174
2.89
-4.22
SW
6
5
56
44
0
38
149
2.47
0.38
MW
232
0
366
728
112
118
1556
25.81
0.06
WA
96
0
282
907
80
52
1417
23.51
-2.53
TF
91
0
489
181
96
59
916
15.20
4.32
CL
173
135
326
502
232
448
1816
30.13
8.83
Total1977
615
140
1540
2479
528
726
6028
PL (%)
10.20
2.32
25.55
41.12
8.76
12.04
100.00
Table 3 Land cover transition matrix in 1977-1994 (ha)
PL: percent of landscape; AACR: annual average changing rate.
2000
1994
RM
SW
MW
WA
TF
CL
Total2000
PL (%)
AACR (%)
RM
SW
MW
WA
TF
CL
Total1994
PL (%)
68
12
27
38
76
15
236
3.92
5.94
20
124
20
1
0
1
166
2.75
1.90
65
13
1079
74
96
265
1592
26.41
0.39
1
0
31
1228
311
6
1577
26.16
1.88
20
0
93
33
430
27
603
10.00
-5.70
0
0
306
43
3
1502
1854
30.76
0.35
174
149
1556
1417
916
1816
6028
2.89
2.47
25.81
23.51
15.20
30.13
Table 4 Land cover transition matrix in 1994-2000 (ha)
249
100.00
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B2. Beijing 2008
2006
2000
RM
SW
MW
WA
TF
CL
Toatl2006
PL (%)
AACR (%)
RM
SW
MW
WA
TF
CL
Total2000
PL (%)
32
0
63
101
62
0
258
4.28
1.55
16
112
14
8
2
0
152
2.52
-2.33
81
28
754
243
135
54
1295
21.48
-3.11
85
13
190
1065
66
46
1465
24.30
-1.18
22
0
74
68
222
39
425
7.05
-4.92
0
13
497
92
116
1715
2433
40.36
5.20
236
166
1592
1577
603
1854
6028
3.92
2.75
26.41
26.16
10.00
30.76
100.00
Table 5 Land cover transition matrix in 2000-2006 (ha)
Year
1977
1996
2000
2006
1977
1996
2000
2006
1977
1996
2000
2006
PD
AREA_MN
LSI
RM
0.4142
0.9281
1.0109
0.4640
24.6132
16.3591
13.1739
14.9496
6.9217
12.4901
12.1111
8.2774
SW
0.1160
0.0994
0.3812
0.0663
20.0443
24.8400
11.7783
38.5875
3.2785
3.1829
7.7545
2.9036
MW
0.4970
0.7624
0.9115
0.8783
41.4030
31.6780
28.9587
26.9609
8.6128
10.5686
11.9438
11.6680
WA
0.1325
0.0663
0.0166
0.0166
346.7362
354.1275
1342.35
1422.99
7.8693
14.8805
13.7429
12.2857
TF
0.2816
0.2320
0.1326
0.2320
31.0712
12.5293
20.7900
18.4114
5.8896
5.1685
3.6860
5.1759
CL
0.2319
0.5138
0.5469
0.4309
52.5921
61.9171
56.3100
90.4431
7.0331
9.1058
9.0521
8.4877
Table 6 Landscape indices of wetlands during 1977-2006
Year
CONTAG
PD
SHDI
SHEI
D
1977
51.6748
1.6733
1.4726
0.8219
0.4173
1996
47.3397
2.6021
1.5282
0.8529
0.3755
2000
46.0629
2.9996
1.5555
0.8681
0.3904
2006
50.3289
2.0881
1.4626
0.8163
0.4858
as that in 1970s. The growth in population reflects on the farm
produce and water consumption. At the same time, the area of
CL increased noticeably from 726ha in 1977 to 2433ha in 2006.
Large areas of MW and WA converted into CL, which caused
in the shrinkage of wetlands and the changes of landscape
pattern.
3.3.2 Economic policy spurred landscape change
Since 1978, China initiated the economic reform and open-door
policy, the GDP of Guide country has clearly increased, and the
area of CL expanded from 1977 to 2000. During the period
1994-2000, the trend of changing wetlands to CL decreased,
which was attributable to the ecology construction policy of the
country in the end of 1980s, such as the conversion of farmland
to grassland. After 2000, the area of CL increased remarkably
again and became the dominant landscape type, the influence of
human beings was a main reason. The shape of CL patches was
regular and the CL extended to connect each other, which
resulted in the increase of dominant index and contagion index
on the whole.
Table 7 Landscape indices of wetlands during 1977-2006
3.3 The reasons analysis for wetlands dynamics
3.3.1
Population growth impulses landscape change
110000
y = 991.04x - 2E+06
Population
100000
2
R = 0.9836
90000
80000
70000
population
60000
50000
1965
1970
1975
1980
1985 1990
Year (a)
1995
2000
2005
2010
Figure3. Population in Guide Country during 1970-2006
Over the past 30 years, the population of Guide country grew
very fast (Figure3). The population in 2000s is about 1.6 times
250
3.3.3 Construction of reservoir results in landscape change
The construction of reservoir greatly impacts the wetland
hydrological balance, and in consequence the landscape pattern
changed remarkably. Guide wetlands situated on the
downstream of Longyangxia reservoir and the upstream of
Lijiaxia reservoir. After the storing water and running of
Longyangxia Hydroelectric station in 1987, the freshwater
source of wetlands was obstructed, and the area of WA
decreased during 1977-1994. Many washes disappeared and the
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B2. Beijing 2008
PD of WA dropped greatly. Almost all the wetland types had
undergone a significant change during the same period. The
Lijiaxia reservoir was constructed in 1996. The freshwater
source of wetlands was supplied in time by the Lijiaxia
reservoir storing water, especially in dry season, which caused
the increase of WA area during 1994 to 2000. Many large areas
of TF converted into WA and other wetland types in this period.
The construction and operation of reservoirs is very important
to irrigation. The area of CL nearby watershed increased
continuously following the reservoir construction.
3.3.4 Climatic information affects landscape change
According to the climatic changes in Guide country during the
period 1970-2006 (Figure4), the mean annual air temperature
was on a ascending trend with a rate of 3.43℃ per decade, and
increased noticeably with a higher rate in the 2000s. The annual
precipitation was highly variable and had no notable change on
the whole. Overall, the climate became warmer and dryer
during the past 30 years, which was one main reason for the
shrinkage of wetlands. During 1994-2000, the annual
precipitation value was lower than the perennial average value,
however, the landscape of wetlands was improved obviously.
This was mainly because that the reservoirs self-function
supplied freshwater source to wetlands in time. It indicated that
the reservoir self-function played a very important role in
maintaining the hydrological balance of wetlands.
Acreman, M.C., Hollis, G.E., 1996. Water Management and
Wetlands in Sub-Saharan Africa. IUCN, Gland, Switzerland.
Antrop, M., 2000. Background concepts for integrated
landscape analysis. Agriculture, Ecosystems and Environment,
77, pp. 17-28.
Alejandro Flamenco-sandoval, 2007. Assessing implications of
land-use and land-cover change dynamics for conservation of a
highly diverse tropical rain forest. Biological conservation, 138,
pp. 131-145.
Baker J R,Briggs S A,Gordon V et al., 1991. Advances in
classification for land cover mapping using SPOTHRV imagery.
International Journal of Remote Sensing, 12, pp. 1071-1085.
Cai X J, 2006. Eco-environmental problems lid ahead of the
Yellow River Delta Wetland ecosystem and their control
countermeasures. Marine environmental science, 25(2), pp.
88-91.
Huang Y J, 2002. Spatial Integrated Analysis on Water Erosion
of Yellow River China. IEEE, pp. 2411-2413.
air temperature
y = -0.101x+ 509.36
2
R = 0.0002
14.0
13.5
13.0
12.5
12.0
11.5
y = 0.0343x- 56.052 11.0
2
10.5
R = 0.3441
10.0
2000
2005
2010
Jiang S Z, 2008. Analysis on variety trend of runoff between
Dari and Maqu in the headwater region of Yellow Rriver in the
past 50 years. Geographical research, 27(1), pp. 221-228.
0
(mm)
REFERENCE
( C)
precipitation
500
450
400
350
300
250
200
150
100
1965
driving force for dynamics of Guide wetlands.
1970
1975
1980
1985
1990
1995
Year(a)
Li X T, 2007. Study on dynamic change of wetlands with in
Yellow River Delta under new water-sediment circum stances.
Water Resources and Hydropower Engineering, 38(11), pp.
18-21.
Figure4. Climatic variations in Guide County during 1970-2006
4. CONCLUSIONS
(1) As a whole, the area of wetlands showed a downward trend
during the past 30 years. The area of wetlands decreased from
5302ha in 1977 to 4212ha in1994, which mainly resulted from
the exquisite decrease of WA and RW area. During 1994-2006,
the shrinkage of wetlands was restrained effectively, whereas
the area of CL was on an ascending trend all the time.
(2) From 1997 to 2000, the wetlands became more fragile and
less connective. The landscape types converted to each other
dramatically, which resulted in the shape of patches inclined to
complicated and irregular. During 2000-2006, the area of CL
expanded more noticeably compared with that during
1994-2000, both the dominant and contagion indices on the
whole area were enhanced.
(3) After analyzing the dynamics of Guide wetlands, the result
shows that many large areas of wetlands converted into CL
during 1977-2006. It revealed that the human beings explored
wetland resource very dramatically. The warming and drying
climate of the country was also the direct reason for the
shrinkage of wetlands. The Longyangxia and Lijiaxia reservoirs,
located in the upper Yellow River, which greatly impacted the
wetland hydrological process and characteristics, was the
251
Lu, L., Li, X., Cheng, G.D., 2003. Landscape evolution in the
middle Heihe River Basin of north-west China during the last
decade. Journal of Arid Environments, 53 (3), pp. 395–408.
Nellis M, 1986. Remote sensing for monitoring rangeland
management strategies in the Kansas Flint Hills.
Proceedings of the Symposium of Commission IV, pp. 370-375.
Rempel, R.S., Carr, A.P., 2003. Patch Analyst Extension for.
ArcView: Version3. Available online at
<http://flash.lakeheadu.ca/~rrempel/patch/index.html>
Sugumaran, R., Harken, J., Gerjevic, J., 2004. Using Remote
Sensing Data to Study Wetland Dynamics in Iowa.
Available from:
<http://www.ia.spacegrant.org/RES_INF/VRR2003/Sugu-SEE
D.pdf>. Site viewed on 26/06/2004.
Xu J X, 2002. Implication of relationships among suspended
sediment size, water discharge and suspended sediment
concentration: the Yellow Rasin, China. Catena, 49, pp.
289-307.
Yoshiki Saitoa, 2000. Delta progradation and chenier formation
in the Huanghe Delta, China. J. of Asian Earth Sciences, 18, pp.
489-497.
Available from: http://www.sciencedirect.com
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B2. Beijing 2008
252